Usability Assessment of Different Contact Models for Modelling the Failure Behaviour of Sedimentary Rocks Under Unconfined Stress Conditions

Abstract

The methods applied in rock material and rock mass studies related to rock mechanics have begun to differentiate with computational and software technology development in recent years. Particle Flow Code software, which is based on the discrete element method, is used in many studies on rock mechanics. With Particle Flow Code, rocks can be modeled in both two and three dimensions, and the behavior of rocks under different conditions can be analyzed. Rocks can be formed using disks in two dimensions and spherical particles in three dimensions. These particles are bonded to each other by contact models with different micro-mechanical properties. The failure of the model occurs as a result of the fracturing with the rupture of these bonds. Using parameters such as density, strength, deformation, which can be defined as the mathematical expression of the natural properties of rock samples, studies can be carried out for different stress conditions through models created after multi-stage calibration steps. In order to have a calibrated and representative model, chosen contact models should be mimic the failure of the rock material. For this purpose, a calibrated model was created for both PBM (Parallel-bonded model) and FJC (Flat-jointed model) using the results of the unconfined compressive strength test of Hawkesbury sandstone. In the calibration phase, the results obtained from the models were matched from the laboratory experiments by using elasticity modulus, uniaxial compressive strength, tensile strength, and Poisson ratio parameters. Results were evaluated by considering the displacement vectors, force chains, and crack resolution parameters. PBM and FJM were similar in terms of displacement vectors and force chains. However, in terms of fracture resolution, FJM stands out as it can be used to analyze failure behavior and modeling studies for sedimentary rocks by providing data up to four times that of PBM.